Triple contrast computed tomography reveals site‐specific biomechanical differences in the human knee joint—A proof of concept study

Cartilage and synovial fluid are challenging to observe separately in native computed tomography (CT). We report the use of triple contrast agent (bismuth nanoparticles [BiNPs], CA4+, and gadoteridol) to image and segment cartilage in cadaveric knee joints with a clinical CT scanner. We hypothesize that BiNPs will remain in synovial fluid while the CA4+ and gadoteridol will diffuse into cartilage, allowing (1) segmentation of cartilage, and (2) evaluation of cartilage biomechanical properties based on contrast agent concentrations. To investigate these hypotheses, triple contrast agent was injected into both knee joints of a cadaver (N = 1), imaged with a clinical CT at multiple timepoints during the contrast agent diffusion. Knee joints were extracted, imaged with micro‐CT (µCT), and biomechanical properties of the cartilage surface were determined by stress‐relaxation mapping. Cartilage was segmented and contrast agent concentrations (CA4+ and gadoteridol) were compared with the biomechanical properties at multiple locations (n = 185). Spearman's correlation between cartilage thickness from clinical CT and reference µCT images verifies successful and reliable segmentation. CA4+ concentration is significantly higher in femoral than in tibial cartilage at 60 min and further timepoints, which corresponds to the higher Young's modulus observed in femoral cartilage. In this pilot study, we show that (1) large BiNPs do not diffuse into cartilage, facilitating straightforward segmentation of human knee joint cartilage in a clinical setting, and (2) CA4+ concentration in cartilage reflects the biomechanical differences between femoral and tibial cartilage. Thus, the triple contrast agent CT shows potential in cartilage morphology and condition estimation in clinical CT.

Articular cartilage and its unique load-bearing properties are essential for the proper function of the knee joint. 1 Cartilage comprises mainly of proteoglycans, collagens, and interstitial fluid, and its mechanical properties are a consequence of its complex structure and composition. 1During instantaneous or cyclic loading, the fluid within the tissue becomes pressurized, and, together with the collagen network, defines the cartilage stiffness.In prolonged loading after the cessation of fluid flow, the equilibrium stiffness of cartilage is attributed to the proteoglycans.Knee injuries cause damage to articular cartilage and can result in posttraumatic osteoarthritis (PTOA). 2 Initial osteoarthritic changes in articular cartilage include microscopic structural and compositional changes.In PTOA, the biomechanical properties of cartilage are compromised due to proteoglycan loss, collagen fibrillation, and increased water content, making the tissue vulnerable to further deterioration. 3An effective, early diagnostic method for simultaneous evaluation of cartilage morphology and composition would allow the opportunity for intervention earlier than currently possible. 2,46][7][8] It requires two subsequent images taken within a 45-120 min wait time: the first image, that is, arthrography, detects cartilage morphology while the second image evaluates the staining of the contrast agents and, consequently, cartilage condition.Since proteoglycans afford a negative fixed charge density in the cartilage, cationic contrast agents are more sensitive to changes in proteoglycan content compared to conventional anionic agents. 9,10CT images obtained using the cationic, iodine-based contrast agent CA4+ 11 reflect the proteoglycan content of the human articular cartilage at diffusion equilibrium. 11The attainment of CA4+ diffusion equilibrium in human knee cartilage ex vivo has been reported to take more than 50 h, 12 which is not feasible in a clinical setting.Two factors mainly contribute to the cationic agent diffusion: (1) the negative charge density from the proteoglycans, that is, higher proteoglycan content results in greater diffusion into the cartilage, and (2) steric hindrance caused by porosity, that is, the flux is greater when the porosity is higher.However, during osteoarthritis progression (fewer proteoglycans and higher porosity), these effects contradict each other.5][16][17] This is feasible because clinical CT can concurrently determine the concentrations of the dual contrast agents (CA4+ and gadoteridol) within cartilage. 14nkanen et al. 20 have introduced a triple contrast agent consisting of CA4+, gadoteridol, and bismuth nanoparticles (BiNP) that, in combination with synchrotron micro-CT (µCT), enabled simultaneous evaluation of articular cartilage morphology and composition. 20The BiNPs (>200 nm) enable distinction between the cartilage surface and synovial fluid, thus, eliminating the need for two subsequent images as required by conventional arthrography methods. 20,21However, previous studies on dual and triple contrast imaging have mainly focused on investigating osteochondral plugs, 12,[14][15][16][17][18][19][20] although ex vivo evaluation of pony joints using dual contrast has been reported more recently. 22This pilot study marks the first use of the triple contrast agent in a clinical CT setting for whole human knee joints, as non-ionic contrast agents and the triple contrast agent have not been employed in this context before.We aim to (1) achieve straightforward and reliable segmentation of articular cartilage via the detection of the synovial fluid-cartilage surface interface, (2) determine CA4+ and gadoteridol concentrations within cartilage, and (3) assess the relationship between CT attenuation and articular cartilage biomechanical properties.

| Contrast agent preparation and CT imaging of cadaveric knee joints
Triple contrast agent solution consisted of phosphate-buffered saline (PBS, 305 mOsm/kg), cationic iodine-based CA4+ (q = +4, M = 1499.9g/mol, 252 mOsm/kg), nonionic gadolinium-based gadoteridol (q = 0, M = 558.7 g/mol, 638 mOsm/kg; ProHance, Bracco International BV) and BiNP (q = 0, M = 209.0g/mol).The concentrations were 6 mg I/mL, 18 mg Gd/mL, and 40 mg Bi/mL for CA4+, gadoteridol, and BiNPs, respectively.The osmolality of CA4+ solution, gadoteridol, and the PBS used for the BiNP suspension were initially determined using a freezing point osmometer (Halbmikro-osmometer; GWB; KNAUER Wissenschaftliche Geräte GmbH).Subsequently, the osmolality of the triple contrast agent was adjusted to 310 mOsm/kg by adding NaCl, taking into account the osmolalities of the individual components, to correspond to saline osmolality and to ensure comparability between imaging and biomechanical experiments.The PBS included protease inhibitors ethylenediamine-tetra-acetic acid disodium salt dihydrate (C = 1.86 g/L; VWR International) and benzamidine hydrochloride hydrate (C = 0.78 g/L; Sigma-Aldrich Co.).CA4+ had been prepared as described previously, 11 while gadoteridol was commercial and purchased.BiNPs were prepared as described in the Supporting Information: Section S1.Additionally, the rheological properties of the triple contrast agent mixed with synovial fluid were studied as described in Supporting Information: Section S2.
A male cadaver (age 67 years) referred to an autopsy in Kuopio University Hospital was obtained for this study.The sample collecting has been approved by the ethical committee of Kuopio University Hospital (decision number 134/2015[58/2013]).Before the experiments, the cadaver was stored in 4°C.CT imaging was conducted in room temperature with a full-body clinical CT scanner (SOMATOM Definition AS; Siemens Healthcare) using two subsequent tube voltages, 70 and 140 kVp.The combined duration of the two subsequent scans was 2 min.Axial images were reconstructed using a bone tissue convolution kernel (Q30s) with in-plane pixel size of 0.38 mm × 0.38 mm, and slice thickness of 0.9 mm.First, a native image without any contrast agent was taken for reference, after which the triple contrast agent (40 mL to each knee to ensure complete fill-up) was injected intra-articularly under the patella (Figure 1A).The knees were extended and flexed manually for 60 s to enable even distribution of contrast agent into the joint cavity, and subsequently, imaged again 3 min after the injection.CT imaging was repeated at 15, 30, 45, 60, 90, 120, 150, and 180 min after the injection.To ensure that the contrast agent would diffuse evenly in the joint, and to prevent the contrast agent from concentrating on the popliteal area of the knee, the knees were also manually flexed F I G U R E 1 (A) Triple contrast agent was injected intra-articularly into both knees of a cadaver, and the knees were imaged using a clinical CT with two energies (70 and 140 kVp) before injection, and at several timepoints after the injection up to 180-min timepoint.(B) The CT images were coregistered, and the cartilage was segmented from the 60-min timepoint images.Contrast agent concentrations in the segmented cartilage were calculated from the attenuation maps at different imaging energies.(C) After knee joint extraction, the obtained osteochondral samples were imaged in the air using µCT.Biomechanical mapping was conducted using indentation stress-relaxation protocol, and the equilibrium and instantaneous Young's moduli were calculated for the measured locations, and maps of the surfaces were created.µCT, micro-CT; CT, computed tomography.
ORAVA ET AL.
| 417 and extended between every imaging as described in the Supporting Information: Section S3.
After CT scanning, contrast agents were rinsed out of the knees with arthroscopic knee washout using PBS until the washout coming out of the knee was clear (visually evaluated), to speed up the following proper wash-out before further experiments.The following day, articulating surfaces of the femurs and tibias of the left and right knee were extracted by a pathologist.Then, the samples were frozen at −20°C in PBS until biomechanical testing.

| Segmentation
The CT images of femurs and tibias were co-registered at each timepoint with the native image (before the injection of contrast agent) with MIM Software (version 7.0.5;MIM Software Inc.).Based on an earlier BiNP study, 22 the cartilage was segmented manually using the 60-min timepoint CT images, while the other timepoints served as additional control timepoints.Segmentation was conducted with open-source software 3D Slicer (version 4.11.20210226;Brigham and Women's Hospital), 23,24 using the Level tracing effect of Segment Editor for each image slice in the coronal plane (Figure 1B).A 3D 5-by-5-by-3 median filter (medfilt function in MATLAB, R2021b; The MathWorks Inc.) was applied to the image stack to reduce noise in the CT images before further analysis.

| Contrast agent concentration analysis
Contrast agent concentration estimation was based on the obtained attenuations in Hounsfield units.To allow this, calibration and validation were performed which are detailed in the Supporting Information: Section S4.[20] The diffusion of contrast agents was studied by fitting a diffusion equation to the data: where C max is the maximum contrast agent concentration, t is the diffusion time and τ is the diffusion time constant (i.e., the time required for the contrast agent to reach 63.2% of the maximum concentration). 25

| µCT imaging and biomechanical mapping
The samples were thawed, and any remaining contrast agent was washed out of the cartilage over a period of 4 days as described in the Supporting Information: Section S5.Articular cartilage thickness was obtained from µCT images (Figure 1C) by using a custom-made MATLAB code.The imaging was conducted in the air with Nikon XT H 225 µCT scanner (Nikon Metrology Europe), with an isotropic voxel size of 50 µm × 50 µm × 50 µm, tube voltage of 120 kVp, tube current of 167 µA, tube power of 20.0 W, and filtering with 0.5 mm aluminum filter.Mapping of the biomechanical properties was conducted using a multiaxial mechanical tester Mach-1 v500css (MA006; Biomomentum Inc.) with a multiaxial 17 N load cell (MA232; Biomomentum Inc.).Details on the stress-relaxation protocol can be found in Supporting Information: Section S5.For each measured location, instantaneous and equilibrium Young's moduli were calculated as described in Supporting Information: Section S5.

| Statistical analysis
The correlation between cartilage thicknesses obtained from clinical

| RESULTS
The tibiofemoral cartilage of the knees, segmented from the 60-min The gadoteridol concentration (Figure 3B) in femoral cartilage was significantly higher than in tibial cartilage at the following  properties based on the diffused contrast agent concentrations.In agreement with previous observations in osteochondral plugs, 17,19 CA4+ concentration is greater in femoral cartilage, which has a higher equilibrium modulus, than in tibial cartilage.As the biomechanical moduli of articular cartilage decrease during the progression of osteoarthritis, 26,27 these changes in the relationship may serve as a potential future diagnostic tool for the disease.
The segmentation of articular cartilage in the current study is straightforward and, based on µCT reference imaging, reliable.Even though we conducted manual segmentation, the BiNPs enable the use of a level tracing tool, which accelerates the segmentation process compared to a completely manual segmentation.The BiNPs provide sustained contrast between the articular cartilage and synovial fluid throughout the 180-min imaging experiment.In the future, semiautomatic 28 or artificial intelligence-based 29 segmentation tools could be used to speed up and ease the segmentation and image analysis.
The obtained contrast agent concentrations reveal biomechanical differences between femoral and tibial cartilage.After 60 min, the CA4+ concentration is consistently higher in the femoral than in the tibial cartilage.1][32] In the current analysis, we subtracted the attenuation of native cartilage from the attenuation maps, before calculating the contrast agent concentrations in the cartilage.We performed this subtraction because the used material decomposition method would otherwise estimate the attenuation caused by the cartilage to be a result of either the gadoteridol or CA4+.However, with this approach, two separate CT scans would be needed, which can be time-consuming and impractical in a clinical setting.As one of the motivations for using the triple contrast agent is to eliminate the first arthrography, we estimated the contrast agent concentrations from the total attenuation maps without removing the native cartilage attenuation.In this way, the significant differences in contrast agent concentrations are not yet observable at 60-min timepoint.However, using the total attenuation maps, the CA4+ concentration in femoral cartilage is significantly higher than in tibial cartilage at 150-and 180-min timepoints.Additionally, the overall findings remain consistent regardless of whether the bulk mean attenuation is subtracted or not.In an earlier study, native cartilage attenuation was successfully subtracted using an approximated bulk value everywhere within the cartilage. 22The triple contrast method thus shows potential in characterizing cartilage condition using a single clinical CT scan.Improved contrast agents with faster diffusion kinetics and higher contrast-to-noise ratios could potentially mitigate the issue.To evaluate the validity of the results obtained from the cadaveric samples we compared the results to earlier studies.In 180 min, the CA4+ concentration reaches 60%-72% of the initial contrast agent concentration (6 mg I/mL), being in the range of the previously reported magnitudes (56%-144%). 12,14,15,17,20In previous studies, using dual contrast agent (CA4+ and gadoteridol) to study human osteochondral plugs, several bath concentrations were explored between 5 and 10 mg I/mL.The CA4+ partition depends nonlinearly on the used contrast agent bath concentration, with lower bath concentration leading to relatively higher partitions. 25an CA4+ partitions of 144% (bath concentration of 6 mg I/mL, clinical CT) 14 and 66% (bath concentration of 5 mg I/mL, synchrotron µCT) 17 are reported with 120-min diffusion time.In the contrast agent diffusion study by Bhattarai et al., 12 the CA4+ partition after 180 min was 56% (bath concentration of 10 mg I/mL). 12Even though the bath concentrations are different in the previous study, the obtained partition is very similar in the current study.In a previous triple contrast agent (CA4+, gadoteridol, and BiNPs) study with bovine osteochondral plugs, the CA4+ partitions after 120 min of diffusion were 89%-137% (bath concentration of 5 mg I/mL). 20The diffusion time constant obtained from the diffusion fits (median 75 min, interquartile range: 97 min) is of the same magnitude as previously reported by Stewart et al. 25 (104 min) in an osteochondral plug experiment at room temperature.However, in the current study, even though the experiments were conducted at room temperature, the temperature of the knee joint was not monitored.
Previously, normalization of CA4+ concentration with gadoteridol concentration diminished the effect of water content on the early diffusion of CA4+. 12,14,15,17However, in the current study, the gadoteridol concentrations are so low that with the normalized CA4+, the differences found with comparing CA4+ concentrations are lost, and the normalization did not improve the results as previously reported 17 and now anticipated.Gadoteridol concentrations in the cartilage remained very low throughout the diffusion, reaching only 6%-8% of the injected contrast agent concentration (18 mg Gd/mL) in 180 min.Additionally, the maximum gadoteridol concentration obtained from the diffusion fit (1.62 mg Gd/mL) is only 9% of the injected concentration.In previous studies using dual contrast agent (CA4+ and gadoteridol), mean gadoteridol partitions were 52% (bath concentration of 18 mg Gd/mL, clinical CT) 14 and 33% (bath concentration of 10 mg Gd/mL, synchrotron µCT) after 120-min diffusion. 17Additionally, Bhattarai et al., 12 using human osteochondral plugs and µCT, report the gadoteridol concentration at 180 min to be 9 mg Gd/mL (bath concentration of 20 mg Gd/mL). 12In bovine osteochondral plugs, gadoteridol partitions were 79%-93% using a triple contrast agent (bath concentration of 10 mg Gd/mL) after 120min diffusion using synchrotron µCT. 20Thus, gadoteridol concentrations in the current study are notably lower than reported in the previous osteochondral plug studies. 12,14,15,17,20To our knowledge, the gadoteridol diffusion in knee joint cartilage surfaces has not been studied earlier.Even though the earlier osteochondral plug studies 12,14,15,17 suggest the normalization with gadoteridol would allow enhancing the sensitivity of CA4+ at early timepoints, the results of the current study do not indicate similar improvement in whole knee joints using clinical CT.Gadoteridol may also diffuse in other surrounding soft tissues instead of cartilage, leaving the gadoteridol concentration in cartilage lower than expected.Moreover, the material decomposition used for the simultaneous determination of CA4+ and gadoteridol concentrations may be affected by the relatively large voxel size, challenging imaging geometry, and beam hardening.However, any potential beam hardening effect was systematic across timepoints and did not impact the calibration and validation contrast agent tubes positioned between the knee joints.
In future studies, higher contrast agent concentrations should be explored to enhance the contrast differences using a single CT scan, especially for gadoteridol.Additionally, in vivo contrast agent diffusion is likely quicker due to higher body temperature.Thus, further research is needed to reveal if higher gadoteridol concentration in synovial fluid enhances the normalization protocol, or if it is needed in the protocol at all.The material decomposition method for determining contrast agent concentrations is based on the k-edges of the elements, here, 33.2 keV for iodine and 50.2 keV for gadolinium.
In future, the method could also be improved by enhancing the element separation by using elements that have their k-edges further from each other, that is, combining CA4+ with a neutral contrast agent based on an element with a higher k-edge, such as tantalum (67.4 keV).Additionally, the triple contrast agent could be combined with a photon-counting detector (PCD) CT, allowing for higher spatial resolution, increased signal-to-noise ratio, radiation dose efficiency, and multiple material separation from a single scan due to simultaneous data acquisition with multiple energies. 33Dual contrast agent (CA4+ and gadoteridol) is suitable for cartilage imaging with an experimental PCD setup, 18 and full-body PCD-CT with a single contrast agent has been used to evaluate stifle joint cartilage damage in swine osteoarthritis model. 34mitations of the study include the following.We only used one cadaver, albeit we did analyze both tibial and femoral cartilages which provides significant structural and compositional variation between the measured points with respect to location, as well as condition, providing a diverse data set for correlation analysis.We conducted biomechanical mapping over the entire joint surfaces, and altogether, 201 biomechanical measurements were obtained.Originally, 256 biomechanical measurements were intended, but the present biomechanical measurement protocol was unable to detect the cartilage surface reliably in all points, especially in the tibial samples, and hence, we were unable to conduct the indentation.This is most probably due to the cartilage being seemingly damaged and appearing to be fibrillated and eroded, which leads to uncertainties in defining the surface contact and, consequently, the orientation of the surface.An additional 16 points had to be excluded from the analysis, as the cartilage was so thin that it was not detected in the clinical CT images.Therefore, there were no contrast agent concentrations to be compared in those locations.Nevertheless, we expected the total of 185 measurement points used in the analysis between biomechanical properties and contrast agent concentrations to be sufficient for finding the correlation.In previous in vitro studies, ORAVA ET AL.
| 421 smaller sample sizes (12-53 samples of 2-4 cadavers) have been sufficient for the correlation between biomechanical properties and contrast agent concentrations for human cartilage tissue. 11,12,17,18though the site-wise difference in CA4+ concentration reflected the biomechanical properties, surprisingly, no correlation could be established between the contrast agent concentrations and biomechanical moduli.Additionally, further research with a diverse sample set (e.g., age, sex, health of cartilage tissue) is needed to verify the generalization of the results obtained with one cadaver in this proof-of-concept study.However, even the observed relative site-specific differences would allow making subject-specific osteoarthritic degeneration models to be more accurate, as subject-specific mechanical and material properties lead to different mechanical responses in subject-specific modeling. 35,36croscopical and spectroscopical characterization of cartilage structure and composition was out of the scope for the current study.
However, the cartilage was visibly worn and fringed, especially in the tibial plateaus and the left medial femur (Figure 4).The obtained cartilage biomechanical properties suggest cartilage damage and decreased proteoglycan content: even though the equilibrium moduli in the present study (ranging from 0.01 to 0.96 MPa) are of the same magnitude as the previously reported values for human cartilage, 12,19,26,27,31 the mean equilibrium moduli of femoral and tibial cartilage (0.33 and 0.21 MPa, respectively), correspond to values previously reported for severely osteoarthritic femoral and tibial cartilage. 26,27As the knee joints were imaged as a whole, biomechanical experiments had to be conducted after contrast agent diffusion.Unfortunately, the effect of isotonic contrast agents on cartilage biomechanics has not been studied to our knowledge.However, as the observed biomechanical moduli fall within the previously reported values, 12,19,26,27,31 we believe the contrast agent diffusion did not affect biomechanics due to the thorough washout of remaining contrast agents before biomechanical experiments.Earlier dual and triple contrast studies utilized µCT 12,13,15,[17][18][19][20] or a clinical CT scanner 14 in an in vitro setup, which allowed for significantly smaller voxel size compared to the present study.Furthermore, this is the first time that contrast agent concentrations, determined with a clinical CT scanner, are compared with cartilage biomechanical properties.In the current study, the voxel size of the clinical CT, especially in the axial plane, is rather large (0.38 mm × 0.38 mm × 0.9 mm) compared to the average cartilage thickness (3.7 mm), meaning the cartilage thickness is around 4 voxels, on average.For this reason, depth-dependent contrast agent concentrations within cartilage are not investigated.Comparably, when clinical CT was previously used in an in vitro study, a significantly smaller voxel size was used, and the mean cartilage thickness was over 20 voxels. 14Additionally, the partial volume effect, that is, the attenuation measured in a single voxel comprising of both cartilage and bone or synovial fluid, might have increased the segmented cartilage attenuation on cartilage-bone and cartilage-synovial fluid interfaces.However, as the cartilage thickness obtained from clinical CT images correlates well with the thickness obtained from the µCT images acquired in air, and the mean difference between the thicknesses is less than half the slice thickness, partial volume effects are considered to have only a minor effect on the segmentation accuracy.
Even though the safety of the triple contrast agent has not yet been studied, the safety of the single contrast agents used has been addressed previously.Gadoteridol is a commercial contrast agent approved for use in the clinic.In preliminary studies, CA4+ shows no toxicity in an in vivo rat model 11 and clears from in vivo rabbit knee joints by 24 h, leaving no signs of inflammation, swelling, or acute toxicity. 25Additionally, BiNPs are biocompatible in vitro and in vivo. 37In the current study, we observed no agglomeration of BiNPs in PBS, during 24 h of observation.Moreover, a previous study reports that the BiNPs remain stable in the triple contrast agent solution for more than 24 h. 20Nevertheless, nanoparticles are new for CT imaging and orthopedic research, and thus, additional research and preclinical studies are needed to assess safety and toxicity. 38The preparation of the triple contrast agent in this study was done on the same day as the imaging, minimizing the possibility of BiNP agglomeration.The viscosity experiment shows the viscosity of the triple contrast agent mixed with synovial fluid not to change during the first 12 h (Supporting Information: Section S2 and Figure S1).
In this study, we investigated the triple contrast agent for imaging of a human knee joint with a clinical CT scanner.Our results support this methodology and show potential as a tool for straightforward segmentation and composition assessment of articular cartilage using a feasible single scan timepoint in a clinical setting (150 min after the injection).However, further studies are still needed to confirm the previously observed in vitro correlation between the CA4+ concentration and biomechanical properties for knee joints when using a clinical CT scanner.The limitations of gadoteridol as a neutral contrast agent for triple contrast CT motivate the investigation of alternative neutral contrast agents with a higher and more stable intake in the cartilage.Nevertheless, as the concentration of CA4+ reflects cartilage biomechanics, the use of CA4+ in combination with BiNPs shows potential for the clinical estimation of articular cartilage morphology and degenerative state.
CT and µCT images in biomechanical measurement locations (n = 229) and the correlation between contrast agent concentrations and biomechanical moduli (n = 185) was studied using Spearman's correlation.The correlation was considered statistically significant when p < 0.05.Additionally, the contrast agent concentrations and the biomechanical moduli in femoral and tibial cartilage were compared using Mann-Whitney U-test.The comparison was conducted in the biomechanical measurement points (n = 185), and the femoral and tibial cartilage were considered different when p < 0.05.The analysis was conducted with MATLAB.

4 |
DISCUSSIONWe applied the triple contrast agent imaging method, for the first time, to image cadaveric human knee joints using clinical CT.Triple contrast agent imaging with clinical CT enables facile and reliable segmentation of articular cartilage and synovial fluid.The method reveals relative differences between femoral and tibial cartilage biomechanicalF I G U R E 2 (A) 3Dillustrations of femoral and tibial cartilage in left and right knee, segmented from CT images at the 60-min timepoint.Different colored segmentations illustrate how the joints were cut for the µCT imaging and biomechanical measurements.(B) Scatter plot of cartilage thickness in biomechanical measurement locations obtained from clinical CT and µCT images.Spearman's rho between clinical CT and µCT was ρ = 0.710 (p < 0.001).Black line shows the least-squares line fit, h h = 0.80 × + 1.11 CT μCT mm.3D, three-dimensional; µCT, micro-CT; CT, computed tomography.F I G U R E 3 (A) CA4+ and (B) gadoteridol concentration in femoral (blue) and tibial cartilage (red) at measured timepoints from 15 to 180-min timepoint.(C) Equilibrium and (D) instantaneous modulus in femoral (blue) and tibial cartilage (red).Stars denote statistically significant differences between femurs and tibias (Mann-Whitney U-test, *p < 0.05, **p < 0.01, ***p < 0.001).

F I G U R E 4
Photographs of the extracted left and right femur and tibia (left), CA4+ concentration at 180-min timepoint in segmented cartilage (middle), and equilibrium modulus map interpolated between the biomechanical measurement locations marked with black dots (right).